Moderne künstliche Intelligenz mit Deep Learning ist bereits heute schon im Einsatz in verschiedenen Anwendungen. Sprachsteuerung von Apple mit Siri, Amazon mit Alexa, autonome Fahrzeuge von Waymo, Tesla, Gesichtserkennung von Facebook sind nur einige bekannte Beispiele aus dem Silicon Valley welche Deep Learning einsetzen. Der Vortrag zeigt auf was wir von der Technologie erwarten können und wie Sie unsere Leben beeinflussen wird.

1. Marc Stampfli, Country Sales Manager, Switzerland, 29.6.18
E-Mail: mstampfli@nvidia.com / LinkedIn: https://ch.linkedin.com/in/marcstampfli
MODERN ARTIFICIAL INTELLIGENCE, ROBOTS AND
AUTONOMOUS CARS – REVOLUTION AND CHANCE?

2. 2
ARTIFICIAL INTELLIGENCE, INTELLIGENT
ROBOTS AND SMART MACHINES
PRESS COVERAGE
2014
2018
1968

3. 3
I AM AI

4. 4
WHAT HAPPENED?

5. 5
• Like the physical universe, the digital universe
is large – by 2020 containing nearly as many
digital bits as there are stars in the universe. It
is doubling in size every two years, and by
2020 the digital universe – the data we create
and copy annually – will reach 40 zettabytes, or
40 trillion gigabytes.
• Every microphone, every camera, every sensor
in general, every IoT device is producing
machine generated data – abundance of data is
the fuel for artificial intelligence.
TODAY DATA IS EVERYWHERE AND DEVICES GENERATE IT
#1- BIG DATA AND REAL WORLD DATA

6. 6
#2 - RE-DISCOVERY OF
ARTIFICIAL NEURONAL NETWORKS
Artificial Intelligence (AI) has been part of our
imaginations and simmering in research labs
since a handful of computer scientists rallied
around the term at the Dartmouth Conferences
in 1956 and birthed the field of AI.
Machine Learning is the practice of using
algorithms to parse data, learn from it, and
then make a determination or prediction about
something in the world rather than hand-coding
software routines.
Artificial Neural Networks, came and mostly
went over the decades. Deep Neural Networks
are inspired by our understanding of the
biology of our brains – all those
interconnections between the neurons.
NEW PARALLEL COMPUTING ALGORITHMS FOR DEEP LEARNING

7. 10
AI & DEEP LEARNING —
THE NEW COMPUTING MODEL
“Software that writes software”
“little girl is eating
piece of cake"
LEARNING
ALGORITHM
“millions of trillions
of FLOPS”

8. 11
2016 – Baidu Deep Speech 2
Superhuman Voice Recognition
2015 – Microsoft ResNet
Superhuman Image Recognition
2017 – Google Neural Machine Translation
Near Human Language Translation
100 ExaFLOPS
8700 Million Parameters
20 ExaFLOPS
300 Million Parameters
7 ExaFLOPS
60 Million Parameters
To Tackle Increasingly Complex Challenges
NEURAL NETWORK COMPLEXITY IS EXPLODING

9. 12
1980 1990 2000 2010 2020
GPU-Computing perf
1.5X per year
1000X
by
2025
#3 - GPU COMPUTING ENABLES INNOVATION
Original data up to the year 2010 collected and plotted by M. Horowitz, F. Labonte, O. Shacham, K.
Olukotun, L. Hammond, and C. Batten New plot and data collected for 2010-2015 by K. Rupp
102
103
104
105
106
107
Single-threaded perf
1.5X per year
1.1X per year
APPLICATIONS
SYSTEMS
ALGORITHMS
CUDA
ARCHITECTURE

10. 13
CPU with
multiple Cores
e.g. 12-20 Cores GPU with n-times
Thousands of Cores
e.g. n x 5120 cores
PARALLEL COMPUTING WITH GPU
• A core in a chip is the processing unit which
receives instructions and performs
calculations
• Clock rate refers to the frequency at which
one core of a multi-core processor is running
• More cores means more calculations per
clock cycle
• CPU optimized for sequential serial
processing of complex orders
• GPU optimized for massive parallel
processing of calculations
What makes a GPU different? Parallelization

11. 14
RISE OF MODERN
ARTIFICIAL
INTELLIGENCE

12. 15
HISTORY OF ACCELERATED COMPUTING PAVED THE
WAY TO MODERN AI WITH DEEP LEARNING
2006 2008 2012 20162010 2014
Fermi: World’s
First HPC GPU
Oak Ridge Deploys World’s
Fastest Supercomputer w/ GPUs
World’s First Atomic
Model of HIV Capsid
GPU-Trained AI Machine
Beats World Champion in Go
Stanford Builds AI
Machine using GPUs
World’s First 3-D Mapping
of Human Genome
CUDA Launched
World’s First GPU
Top500 System
Google Outperforms
Humans in ImageNet
Discovered How H1N1
Mutates to Resist Drugs
AlexNet beats expert code
by huge margin using GPUs
2. Deep learning
network with
12 GPUs, that
required
2000 CPUs
1. Deep learning
network beats
human expert
coding
3. Deep learning
network with GPU
outperforms human

13. 16
The big bang of modern AI set off a string of
“superhuman” achievements. In 2015, Google
and Microsoft both beat the best human score
in the ImageNet challenge. DeepMind’s AlphaGo
recorded its historic win over Go champion Lee
Sedol in 2016 and, more recently, beat the best
player in the world, Ke Jie. Breakthroughs in AI
happen almost every day.
AI ACHIEVES
“SUPERHUMAN” RESULTS

14. 17
TEN YEARS OF GPU COMPUTING PAVED THE
WAY TO MODERN AI WITH DEEP LEARNING
2006 2008 2012 20162010 2014
Fermi: World’s
First HPC GPU
Oak Ridge Deploys World’s
Fastest Supercomputer w/ GPUs
World’s First Atomic
Model of HIV Capsid
GPU-Trained AI Machine
Beats World Champion in Go
Stanford Builds AI
Machine using GPUs
World’s First 3-D Mapping
of Human Genome
CUDA Launched
World’s First GPU
Top500 System
Google Outperforms
Humans in ImageNet
Discovered How H1N1
Mutates to Resist Drugs
AlexNet beats expert code
by huge margin using GPUs
Cambrian
Explosion sparked
by
Deep Learning
+
GPU Acceleration

15. 18
THE BIG BANG OF AI
AND THE EXPANDING
UNIVERSE

16. 19

17. 20
GPU computing is the most productive and
pervasive platform for deep learning and AI.
It begins with the most advanced GPUs and
the systems and software we build on top of
them. We integrate and optimize every deep
learning framework. We work with the major
systems companies and every major cloud
service provider to make GPUs available in
data centers and in the cloud. And we create
computers and software to bring AI to edge
devices, such as self-driving cars and
autonomous robots.
NVIDIA POWERING
THE AI REVOLUTION
2
2

18. 21
EMERGENCE AI SUPERCOMPUTERS
AI WORKSTATIONCLOUD-SCALE AI AI DATA CENTER
Cloud platform with the highest
deep learning efficiency
NVIDIA GPU Cloud
The Essential
Instrument for AI
Research
DGX-1
with
Tesla V100 32GB
The Personal
AI Supercomputer
DGX Station
with
Tesla V100 32GB
The World’s Most Powerful
AI System for the Most
Complex AI Challenges
DGX-2
with
Tesla V100 32GB

19. 22
COMMON SOFTWARE STACK ACROSS DGX FAMILY
GPU optimized and accelerated containers
Cloud Service
Provider
• Single, unified stack for deep learning frameworks
• Predictable execution across platforms
• Pervasive reach
DGX Station DGX-1
NVIDIA
GPU Cloud
DGX-2
22

20. 23
MODERN AI
IN ACTION

21. 24
50% Reduction in Emergency
Road Repair Costs
>$6M / Year Savings and
Reduced Risk of Outage
INFRASTRUCTUREHEALTHCARE IOT
AI TO TRANSFORM EVERY INDUSTRY
>80% Accuracy & Immediate
Alert to Radiologists

22. 25
Driving is a learned behavior that people do
as second nature. Yet one that is impossible
to program a computer to perform. Using all
of the AI capabilities of NVIDIA DRIVE PX 2,
our research AI car, BB8, watches humans
drive, and has learned to drive in all kinds
of conditions — on highways and dirt roads,
through obstacle courses, at night, and in the
rain. Processing data from multiple cameras,
BB8 can even look both ways before safely
crossing a busy road on its own.
NVIDIA BB8 AI CAR —
LEARNING BY EXAMPLE
Center CameraLeft Camera Right Camera

23. 26
MOVIE: EXAMPLE SELF DRIVING CARS (LEVEL 5)

24. 27
AI is transforming the spectrum of healthcare,
from detection to diagnosis to treatment. GE
Healthcare has reinvented the echocardiogram
machine by embedding GPU-powered AI in its
Vivid E95 system. Mayo Clinic used GPU-powered
deep learning to discover that genomic data
can be found in MRIs, hidden from traditional
analysis methods.
NVIDIA is teaming up with the National Cancer
Institute, the U.S. Department of Energy, and
several national labs on the “Cancer Moonshot”
to deliver a decade of advances in cancer
prevention, diagnosis, and treatment in just five
years.
THE BRAIN OF
AI HEALTHCARE

25. 28
MOVIE: LIVE RECONSTRUCTION OF 3D MODEL OF HEART WITH ULTRA SONIC

26. 29
Deep learning and affordable sensors have
created the conditions for a Cambrian explosion
of autonomous machines — IoT with AI. NVIDIA
Jetson™ TX2, an embedded AI supercomputer,
delivers 1 TeraFLOPS of performance in a credit
card-sized module. Such power will enable a
new wave of automation in manufacturing,
drones that can inspect hazardous places, and
robots that can deliver the millions of packages
shipped every day.
THE BRAIN OF INTELLIGENT
MACHINES, ROBOTS, DRONES
& IoT

27. 30
ISAAC — ACCELERATED
LEARNING FOR A WORLD OF
INTELLIGENT MACHINES
The Isaac robot simulator, an AI-based software
platform, lets developers train robots in highly
realistic, physics-based virtual environments and
then transfer that knowledge to real-world units.
Developers can set up extensive test scenarios
using deep learning training, and then simulate
them in minutes instead of months.

28. 31
MOVIE: CONVERGING ARTITIFIAL AND REAL WORLD

29. 32
WHAT CAN I DO?

30. 33
UNDERSTANDING MODERN AI WITH DEEP LEARNING
Sign up for the NVIDIA Developer Program
at https://developer.nvidia.com/join
Discover GPU accelerated containers
Start researching using an AI appliance
with NVIDIA DGX Systems
JOIN OUR COMMUNITY
Get in contact with us: mstampfli@nvidia.com
Take a self-paced course online at
www.nvidia.com/dlilabs
View upcoming events or request a
workshop at www.nvidia.com/dli
GET THE BASICS SKILLS
Visit GTC in Munich October 9-11, 2018
BUILD & TRAIN DATA SCIENCE TEAM WITH
DEEP LEARNING KNOWLEDGE
Watch “Deep Learning Demystified”
Listen to the NVIDIA AI Podcast
Review examples of AI in action
Register with 25% discount code «NVMSTAMPFLI»
https://www.gputechconf.eu/

31. 34
> Founded in 1993
> Jensen Huang, Founder & CEO
> 12’000+ employees
> $9.7B in FY18
“World’s Best Performing CEOs”
— Harvard Business Review
“World’s Most Admired Companies”
— Fortune
“World’s Best CEOs”
— Barron’s
“Most Innovative Companies”
— Fast Company
“Employees’ Choice: Highest Rated CEOs”
— Glassdoor
“50 Smartest Companies”
— MIT Tech Review